Platform or Best of Breed? Last Week’s CX AI Deals Suggest Both

Within four days last week, Salesforce agreed to acquire Cresta, Dialpad announced a merger with Ada, and Five9 announced that it’s buying Level AI. Each of these deals folds conversational AI specialists into broader customer experience (CX) platforms and puts a perennial question squarely in front of CX leaders: Should AI agents come from a unified platform or a best-of-breed specialist? Metrigy’s 2026 AI and CX research studies suggest enterprises have already answered, and the answer is, “Both.” 

In our Contact Center & Beyond: Elevating CX 2026-27 study, conducted in the third quarter, 57.2% of the 626 participating companies see value in buying as many applications as possible from a single provider. Yet our AI’s Role in Customer Experience: 2026-27 study, conducted in the second quarter with 759 companies, shows an average of 12 vendors with AI capabilities in place across CX stacks. The companies getting the most from AI don’t resolve that tension so much as exploit it: 76.5% of the  Contact Center & Beyond study’s Success Group, drawn from companies reporting above-average gains in key business measures from AI’s use, see value in a single provider, while 62.9% of the most successful companies  in our AI’s Role in CX study plan to add AI vendors in the next year.

A Week of Deals: The Specialists Fold Into the Platforms

The terms of the deals tell their own story. Salesforce’s deal for Cresta, which offers voice agents, real-time agent assist, and conversation intelligence, comes at a reported $2.7 billion; Five9 will pay $325 million in cash for Level AI, whose AI agents, agent assist, and quality management run on a shared conversation-intelligence layer; and the all-share Dialpad-Ada combination reportedly values Ada at roughly $430 million.

None of these deals arrives in a vacuum. NiCE closed its acquisition of Cognigy, an enterprise conversational and agentic AI provider, in September 2025 in a deal reported at about $955 million; Genesys bought agent builder Pinkfish in June 2026; and Salesforce struck a reported $3.6 billion deal for Fin, the AI customer service business formerly known as Intercom, that same month. The pattern is hard to miss: The conversational AI software, long the technology that CX leaders bolted on top of a contact center or customer relationship management (CRM) system, is becoming part and parcel of the platform itself.

Read the fine print, though, and the acquirers are hedging. Salesforce says Cresta “will continue to support the leading contact center and telephony platforms its customers rely on today”; Five9 will sell Level AI both inside its platform and as a standalone product; and NiCE positioned Cognigy the same way, as part of the CXone platform and as an independent offering.

An Old Question, Recast for AI Agents

Consolidation complicates the sourcing decision rather than settling it. A company that chose Cresta, Level AI, or Ada for its focused expertise now finds that specialist sitting inside a platform it may or may not use, with roadmap priorities that will inevitably tilt toward the new parent. Conversely, a company that standardized on a unified platform that brings together CRM, contact center, knowledge, and AI natively gains capabilities it might otherwise have had to integrate on its own.

Metrigy’s research offers a closer look at how CX leaders are navigating the decision and why those who are seeing the strongest results from AI decline to pick a side.

The Data: Platform Instincts, Multivendor Reality

The platform instinct runs deeper than a general preference for fewer providers. Participants in our Contact Center & Beyond study would like, on average, 48.4% of their tech stack to come from one provider, and when asked to rank which components matter most to source natively from the same provider, they placed AI agents second, behind only CRM and ahead of knowledge management. Additionally, 67.0% say the coordination of AI, humans, data, and workflows to determine the next-best action in a customer journey, otherwise known as “AI orchestration,” belongs in the core of a platform rather than at its edge.

The multivendor reality runs just as deep. In our AI’s Role in CX study, 46.3% of companies plan to increase the number of AI vendors they use in the next 12 months, and they buy AI agents from one or more of a range of provider types. This includes CRM providers (68.9%), contact center as a service (CCaaS) providers (38.5%), and AI specialists (28.5%)—precisely the three vendor types the recent deals are blurring together.

Our CX Assurance for AI-Powered Contact Centers: 2026-27 study, conducted in the third quarter with 428 companies, shows how this plays out operationally: 47.8% of companies assure their AI agents with a mix of native platform tools and a centralized tool, 25.3% use one platform-agnostic tool, and 19.1% rely on each platform’s native tools in silos.

Success Group data is telling, too. In our Contact Center & Beyond study, 76.5% of the Success Group see value in a single provider, as noted, vs. 62.7% of the Non-Success Group; in our AI’s Role in CX study, 62.9% of the Success Group plan to add AI vendors, as noted, vs. 38.8% of the Non-Success Group. Top performers, in other words, want a strong core platform and are still willing to go outside it.

Weighing the Tradeoffs: Neither Path Comes Free

The unified platform’s appeal is real, and our data shows where it comes from:

• Shared data and context – Data fragmentation is the biggest cause of AI project failure identified in additional Metrigy research, our second-quarter AI Organizational Best Practices: 2026-27 study of 759 companies, and a single platform reduces the hand-off points where customer context could get lost.
• Simpler orchestration and accountability – With orchestration built into the core, one provider owns the handoff between AI and human agents, and one contract owns the outcome.
• Fewer integrations to assure – Every additional platform is another environment to test and monitor, and data and visibility into what AI is doing is already the top capability gap (36.8%) noted in our CX Assurance study.

On the other hand, the platform route has costs, and last week’s deals make them hard to ignore:

• Lock-in and roadmap risk – A specialist absorbed into a platform will, sooner or later, prioritize that platform’s customers, and promises of continued support for rival contact centers are worth getting in writing.
• Integration lag – Acquired capabilities rarely land as a clean swap, so buyers should separate what they can deploy today from what depends on future product integration.
• Innovation pace – Technology innovation (58.9%) and the need for better AI reasoning (52.8%) top the reasons companies plan to change their primary customer interaction platform, and 43.5% expect to change provider types within two years, as we found in the Contact Center & Beyond study.

Best of breed offers the opposite tradeoffs. Specialists have historically moved faster on reasoning, voice, and conversation intelligence, which is precisely why platforms are paying to acquire them. Additionally, our Contact Center & Beyond research shows AI specialists rising from the primary customer interaction platform for 27.7% of companies today to 48.1% by the end of 2028. The cost is sprawl. A dozen AI vendors means a dozen data models, pricing schemes, and governance regimes, and with 44.3% of companies in our CX Assurance study experiencing an AI incident at least monthly, failure to escalate to a human (45.2%) chief among them, a gap between two vendors quickly becomes a customer’s problem.

Standards such as the Model Context Protocol (MCP), which lets AI agents connect to applications and data without custom integrations, promise to ease that burden—and Genesys’s acquisition of Pinkfish, whose prebuilt MCP-based tools connect AI agents to common enterprise applications, signals that platform providers see as much. Adoption is still early, however: Fewer than 20% of companies in our Contact Center & Beyond study use MCP to connect CX applications today, though 37.2% plan to by 2027.

The Answer Is Both—With Conditions

The acquirers themselves seem to concede that the either/or framing is a false choice. By keeping Cresta, Level AI, and Cognigy available to customers of competing platforms, they’re betting that buyers want a strong platform core with an open door. That is, not coincidentally, what top performers in our research are already doing. 

For CX leaders whose AI agent provider just changed hands, or might soon, the questions worth asking now include:

• Will the acquired product remain available, supported, and priced competitively for customers that don’t use the new parent’s platform and, if so, for how long?
• Which capabilities ship today, and which depend on integration work yet to be done?
• Can AI agents, wherever they’re sourced, share customer context, knowledge, and escalation logic with the rest of the stack?
• Is there a platform-agnostic way to test, monitor, and audit every AI agent in production, regardless of vendor?
• If the platform’s roadmap diverges from the company’s needs, what does an exit cost?

Last week’s deals won’t be the last, and CX leaders who treat consolidation as a reason to revisit their sourcing strategy will be better positioned for the next one, choosing a core platform for shared data and orchestration while reserving the right to bring in specialists where they demonstrably outperform, and holding every AI agent, from every source, to the same standard of assurance.

Author(s)

Beth Schultz